Abstract
This paper proposes a less-disturbed ecological driving strategy for connected and automated vehicles (CAVs). The proposed strategy integrates offline planning and online tracking. In offline planning, an energy-efficient reference speed is created based on traffic information (such as the average traffic speed) and vehicle characteristics (such as the engine efficiency map) via dynamic programming. The consideration of average traffic speed in speed planning avoids "selfish" optimizations. In online tracking, model predictive control (MPC) is employed to update the vehicle speed in real-time to track the reference speed. A key challenge in applying ecological driving strategies in real driving is that the vehicle must consider other traffic participants. Therefore, this paper combines both longitudinal and lateral control to achieve better speed tracking by overtaking preceding vehicles when necessary. The proposed strategy has been evaluated in simulations in both single road segment scenarios and real traffic environments. Results demonstrate that the less-disturbed eco-driving strategy improves fuel efficiency by 4.53% compared to human drivers.
Introduction
Most ecological driving (eco-driving) strategies suffer from a "selfish optimization" problem. They calculate a theoretically optimal speed profile based on road geometry and signal phase and timing (SPaT), but these plans often fail in real-world traffic. When a CAV encounters a slower vehicle, standard systems typically revert to Adaptive Cruise Control (ACC), slowing down and deviating from the fuel-efficient plan. This "disturbance" significantly reduces the actual energy savings achieved.
The challenge lies in creating a system that doesn't just react to the car in front but actively works to maintain its optimized "green" path. This requires a strategy that integrates high-level traffic flow intelligence with tactical maneuvering capabilities.
The Proposed Hierarchical Control Strategy
The researchers developed a two-stage hierarchical control structure to address these challenges, combining long-term strategic planning with real-time tactical execution.
1. Offline Planning (The Strategist)
Before the trip begins, the system performs "Offline Planning" using Dynamic Programming (DP). It processes data from Infrastructure-to-Vehicle (I2V) communications, including SPaT data from traffic lights and average traffic speeds. The goal is to generate a reference speed profile that minimizes fuel consumption over the entire route. Crucially, the cost function includes a penalty for deviating from the average traffic speed. This ensures the vehicle's behavior is harmonious with the surrounding flow, avoiding optimizations that might cause traffic waves for other drivers.
2. Online Tracking (The Tactical Driver)
During the drive, the "Online Tracking" module uses Model Predictive Control (MPC) to follow the reference speed. Unlike traditional longitudinal-only systems, this strategy incorporates lateral control (steering). A "Motion Decision Making" layer evaluates surrounding traffic using perception sensors. If a preceding vehicle is moving slower than the optimized reference speed, the CAV determines if it is safe and efficient to change lanes and overtake. This capability allows the vehicle to remain "less disturbed" by slower traffic, sticking closer to its energy-efficient plan.
Experimental Evaluation
The strategy was evaluated using high-fidelity simulations and real-world road parameters. The vehicle model was based on a 2.0L gasoline engine powertrain simulated in Autonomie. The test route was a 3.45km stretch of the A512 road in the UK, featuring four signalized intersections. To establish a baseline, data was collected from ten real-world driving tests involving both experienced and inexperienced human drivers on the same route.
Key Results and Insights
The "less-disturbed" strategy demonstrated significant improvements over both human drivers and standard eco-driving benchmarks.
- Fuel Efficiency Gains: The proposed strategy achieved a 4.53% improvement in fuel efficiency compared to human drivers. When compared to existing eco-driving benchmarks that lacked lane-changing capabilities, it reduced fuel consumption by between 3.1% and 8.7%.
- Tracking Precision: The ability to overtake slower vehicles resulted in much higher tracking accuracy. The Root Mean Square Error (RMSE) for speed tracking was only 0.293 m/s, whereas benchmarks without lateral control exceeded 2.5 m/s.
- Lateral Control as an Efficiency Tool: A key insight is that lane changing, often viewed as a convenience feature, is a powerful tool for energy saving. By avoiding the "stop-and-go" cycles forced by slower traffic, the CAV can operate in its most efficient engine regions for longer periods.
Conclusion
The "less-disturbed" eco-driving strategy proves that integrating lateral maneuvering into speed tracking is essential for realizing the full energy-saving potential of CAVs in mixed traffic. While the system currently relies on robust V2X communications, future work will focus on integrating data-driven models to better predict human driver behavior, further refining the vehicle's ability to maintain its optimal "green" trajectory.
Summary updated